www/extensions/compresseur/lib/JavascriptPacker/class.JavaScriptPacker.php

   1 <?php
   2 /* 7 December 2006. version 1.0
   3  *
   4  * This is the php version of the Dean Edwards JavaScript 's Packer,
   5  * Based on :
   6  *
   7  * ParseMaster, version 1.0.2 (2005-08-19) Copyright 2005, Dean Edwards
   8  * a multi-pattern parser.
   9  * KNOWN BUG: erroneous behavior when using escapeChar with a replacement
  10  * value that is a function
  11  *
  12  * packer, version 2.0.2 (2005-08-19) Copyright 2004-2005, Dean Edwards
  13  *
  14  * License: http://creativecommons.org/licenses/LGPL/2.1/
  15  *
  16  * Ported to PHP by Nicolas Martin.
  17  * modified by Mark Fabrizio Jr. to work with php 4
  18  *
  19  * ----------------------------------------------------------------------
  20  *
  21  * examples of usage :
  22  * $myPacker = new JavaScriptPacker($script, 62, true, false);
  23  * $packed = $myPacker->pack();
  24  *
  25  * or
  26  *
  27  * $myPacker = new JavaScriptPacker($script, 'Normal', true, false);
  28  * $packed = $myPacker->pack();
  29  *
  30  * or (default values)
  31  *
  32  * $myPacker = new JavaScriptPacker($script);
  33  * $packed = $myPacker->pack();
  34  *
  35  *
  36  * params of the constructor :
  37  * $script:       the JavaScript to pack, string.
  38  * $encoding:     level of encoding, int or string :
  39  *                0,10,62,95 or 'None', 'Numeric', 'Normal', 'High ASCII'.
  40  *                default: 62.
  41  * $fastDecode:   include the fast decoder in the packed result, boolean.
  42  *                default : true.
  43  * $specialChars: if you are flagged your private and local variables
  44  *                in the script, boolean.
  45  *                default: false.
  46  *
  47  * The pack() method return the compressed JavasScript, as a string.
  48  *
  49  * see http://dean.edwards.name/packer/usage/ for more information.
  50  *
  51  * Notes :
  52  * # [del]need PHP 5 . Tested with PHP 5.1.2[/del]
  53  *   this is a modified version for PHP 4
  54  *
  55  * # The packed result may be different than with the Dean Edwards
  56  *   version, but with the same length. The reason is that the PHP
  57  *   function usort to sort array don't necessarily preserve the
  58  *   original order of two equal member. The Javascript sort function
  59  *   in fact preserve this order (but that's not require by the
  60  *   ECMAScript standard). So the encoded keywords order can be
  61  *   different in the two results.
  62  *
  63  * # Be careful with the 'High ASCII' Level encoding if you use
  64  *   UTF-8 in your files...
  65  */
  66
  67  /*
  68  * modified by Mark Fabrizio Jr. to work with php 4
  69  */
  70
  71
  72 class JavaScriptPacker {
  73         var $IGNORE = '$1';
  74
  75         // validate parameters
  76         var $_script = '';
  77         var $_encoding = 62;
  78         var $_fastDecode = true;
  79         var $_specialChars = false;
  80
  81         var $LITERAL_ENCODING = array(
  82                 'None' => 0,
  83                 'Numeric' => 10,
  84                 'Normal' => 62,
  85                 'High ASCII' => 95
  86         );
  87
  88 // http://doc.spip.org/@JavaScriptPacker
  89         function JavaScriptPacker($_script, $_encoding = 62, $_fastDecode = true, $_specialChars = false)
  90         {
  91                 $this->_script = $_script . "\n";
  92                 if (array_key_exists($_encoding, $this->LITERAL_ENCODING))
  93                         $_encoding = $this->LITERAL_ENCODING[$_encoding];
  94                 $this->_encoding = min((int)$_encoding, 95);
  95                 $this->_fastDecode = $_fastDecode;
  96                 $this->_specialChars = $_specialChars;
  97         }
  98
  99 // http://doc.spip.org/@pack
 100         function pack() {
 101                 $this->_addParser('_basicCompression');
 102                 if ($this->_specialChars)
 103                         $this->_addParser('_encodeSpecialChars');
 104                 if ($this->_encoding)
 105                         $this->_addParser('_encodeKeywords');
 106
 107                 // go!
 108                 return $this->_pack($this->_script);
 109         }
 110
 111         // apply all parsing routines
 112 // http://doc.spip.org/@_pack
 113         function _pack($script) {
 114                 for ($i = 0; isset($this->_parsers[$i]); $i++) {
 115                         $script = call_user_func(array(&$this,$this->_parsers[$i]), $script);
 116                 }
 117                 return $script;
 118         }
 119
 120         // keep a list of parsing functions, they'll be executed all at once
 121         var $_parsers = array();
 122 // http://doc.spip.org/@_addParser
 123         function _addParser($parser) {
 124                 $this->_parsers[] = $parser;
 125         }
 126
 127         // zero encoding - just removal of white space and comments
 128 // http://doc.spip.org/@_basicCompression
 129         function _basicCompression($script) {
 130                 $parser = new ParseMaster();
 131                 // make safe
 132                 $parser->escapeChar = '\\';
 133                 // protect strings
 134                 $parser->add('/\'[^\'\\n\\r]*\'/',$this->IGNORE);
 135                 $parser->add('/"[^"\\n\\r]*"/', $this->IGNORE);
 136                 // remove comments
 137                 $parser->add('/\\/\\/[^\\n\\r]*[\\n\\r]/', "\n");
 138                 $parser->add('/\\/\\*[^*]*\\*+([^\\/][^*]*\\*+)*\\//', ' ');
 139                 // protect regular expressions
 140                 $parser->add('/\\s+(\\/[^\\/\\n\\r\\*][^\\/\\n\\r]*\\/g?i?)/', '$2'); // IGNORE
 141                 $parser->add('/[^\\w\\x24\\/\'"*)\\?:]\\/[^\\/\\n\\r\\*][^\\/\\n\\r]*\\/g?i?/', $this->IGNORE);
 142                 // remove: ;;; doSomething();
 143                 if ($this->_specialChars) $parser->add('/;;;[^\\n\\r]+[\\n\\r]/');
 144                 // remove redundant semi-colons
 145                 $parser->add('/\\(;;\\)/', $this->IGNORE); // protect for (;;) loops
 146                 $parser->add('/;+\\s*([};])/', '$2');
 147                 // apply the above
 148                 $script = $parser->exec($script);
 149
 150                 // remove white-space
 151 #               $parser->add('/(\\b|\\x24)\\s+(\\b|\\x24)/', '$2 $3');
 152 #               $parser->add('/([+\\-])\\s+([+\\-])/', '$2 $3');
 153 #               $parser->add('/\\s+/', '');
 154 # Modif fil@rezo.net pour conserver les \n
 155                 $parser->add('/(\\b|\\x24)[\\t ]+(\\b|\\x24)/', '$2 $3');
 156                 $parser->add('/([+\\-])[\\t ]+([+\\-])/', '$2 $3');
 157                 $parser->add('/[\\t ]+/', '');
 158                 $parser->add('/\\s+/', "\n");
 159                 // done
 160                 return $parser->exec($script);
 161         }
 162
 163 // http://doc.spip.org/@_encodeSpecialChars
 164         function _encodeSpecialChars($script) {
 165                 $parser = new ParseMaster();
 166                 // replace: $name -> n, $$name -> na
 167                 $parser->add('/((\\x24+)([a-zA-Z$_]+))(\\d*)/',
 168                                          array('fn' => '_replace_name')
 169                 );
 170                 // replace: _name -> _0, double-underscore (__name) is ignored
 171                 $regexp = '/\\b_[A-Za-z\\d]\\w*/';
 172                 // build the word list
 173                 $keywords = $this->_analyze($script, $regexp, '_encodePrivate');
 174                 // quick ref
 175                 $encoded = $keywords['encoded'];
 176
 177                 $parser->add($regexp,
 178                         array(
 179                                 'fn' => '_replace_encoded',
 180                                 'data' => $encoded
 181                         )
 182                 );
 183                 return $parser->exec($script);
 184         }
 185
 186 // http://doc.spip.org/@_encodeKeywords
 187         function _encodeKeywords($script) {
 188                 // escape high-ascii values already in the script (i.e. in strings)
 189                 if ($this->_encoding > 62)
 190                         $script = $this->_escape95($script);
 191                 // create the parser
 192                 $parser = new ParseMaster();
 193                 $encode = $this->_getEncoder($this->_encoding);
 194                 // for high-ascii, don't encode single character low-ascii
 195                 $regexp = ($this->_encoding > 62) ? '/\\w\\w+/' : '/\\w+/';
 196                 // build the word list
 197                 $keywords = $this->_analyze($script, $regexp, $encode);
 198                 $encoded = $keywords['encoded'];
 199
 200                 // encode
 201                 $parser->add($regexp,
 202                         array(
 203                                 'fn' => '_replace_encoded',
 204                                 'data' => $encoded
 205                         )
 206                 );
 207                 if (empty($script)) return $script;
 208                 else {
 209                         //$res = $parser->exec($script);
 210                         //$res = $this->_bootStrap($res, $keywords);
 211                         //return $res;
 212                         return $this->_bootStrap($parser->exec($script), $keywords);
 213                 }
 214         }
 215
 216 // http://doc.spip.org/@_analyze
 217         function _analyze($script, $regexp, $encode) {
 218                 // analyse
 219                 // retreive all words in the script
 220                 $all = array();
 221                 preg_match_all($regexp, $script, $all);
 222                 $_sorted = array(); // list of words sorted by frequency
 223                 $_encoded = array(); // dictionary of word->encoding
 224                 $_protected = array(); // instances of "protected" words
 225                 $all = $all[0]; // simulate the javascript comportement of global match
 226                 if (!empty($all)) {
 227                         $unsorted = array(); // same list, not sorted
 228                         $protected = array(); // "protected" words (dictionary of word->"word")
 229                         $value = array(); // dictionary of charCode->encoding (eg. 256->ff)
 230                         $this->_count = array(); // word->count
 231                         $i = count($all); $j = 0; //$word = null;
 232                         // count the occurrences - used for sorting later
 233                         do {
 234                                 --$i;
 235                                 $word = '$' . $all[$i];
 236                                 if (!isset($this->_count[$word])) {
 237                                         $this->_count[$word] = 0;
 238                                         $unsorted[$j] = $word;
 239                                         // make a dictionary of all of the protected words in this script
 240                                         //  these are words that might be mistaken for encoding
 241                                         //if (is_string($encode) && method_exists($this, $encode))
 242                                         $values[$j] = call_user_func(array(&$this, $encode), $j);
 243                                         $protected['$' . $values[$j]] = $j++;
 244                                 }
 245                                 // increment the word counter
 246                                 $this->_count[$word]++;
 247                         } while ($i > 0);
 248                         // prepare to sort the word list, first we must protect
 249                         //  words that are also used as codes. we assign them a code
 250                         //  equivalent to the word itself.
 251                         // e.g. if "do" falls within our encoding range
 252                         //      then we store keywords["do"] = "do";
 253                         // this avoids problems when decoding
 254                         $i = count($unsorted);
 255                         do {
 256                                 $word = $unsorted[--$i];
 257                                 if (isset($protected[$word]) /*!= null*/) {
 258                                         $_sorted[$protected[$word]] = substr($word, 1);
 259                                         $_protected[$protected[$word]] = true;
 260                                         $this->_count[$word] = 0;
 261                                 }
 262                         } while ($i);
 263
 264                         // sort the words by frequency
 265                         // Note: the javascript and php version of sort can be different :
 266                         // in php manual, usort :
 267                         // " If two members compare as equal,
 268                         // their order in the sorted array is undefined."
 269                         // so the final packed script is different of the Dean's javascript version
 270                         // but equivalent.
 271                         // the ECMAscript standard does not guarantee this behaviour,
 272                         // and thus not all browsers (e.g. Mozilla versions dating back to at
 273                         // least 2003) respect this.
 274                         usort($unsorted, array(&$this, '_sortWords'));
 275                         $j = 0;
 276                         // because there are "protected" words in the list
 277                         //  we must add the sorted words around them
 278                         do {
 279                                 if (!isset($_sorted[$i]))
 280                                         $_sorted[$i] = substr($unsorted[$j++], 1);
 281                                 $_encoded[$_sorted[$i]] = $values[$i];
 282                         } while (++$i < count($unsorted));
 283                 }
 284                 return array(
 285                         'sorted'  => $_sorted,
 286                         'encoded' => $_encoded,
 287                         'protected' => $_protected);
 288         }
 289
 290         var $_count = array();
 291 // http://doc.spip.org/@_sortWords
 292         function _sortWords($match1, $match2) {
 293                 return $this->_count[$match2] - $this->_count[$match1];
 294         }
 295
 296         // build the boot function used for loading and decoding
 297 // http://doc.spip.org/@_bootStrap
 298         function _bootStrap($packed, $keywords) {
 299                 $ENCODE = $this->_safeRegExp('$encode\\($count\\)');
 300
 301                 // $packed: the packed script
 302                 $packed = "'" . $this->_escape($packed) . "'";
 303
 304                 // $ascii: base for encoding
 305                 $ascii = min(count($keywords['sorted']), $this->_encoding);
 306                 if ($ascii == 0) $ascii = 1;
 307
 308                 // $count: number of words contained in the script
 309                 $count = count($keywords['sorted']);
 310
 311                 // $keywords: list of words contained in the script
 312                 foreach ($keywords['protected'] as $i=>$value) {
 313                         $keywords['sorted'][$i] = '';
 314                 }
 315                 // convert from a string to an array
 316                 ksort($keywords['sorted']);
 317                 $keywords = "'" . implode('|',$keywords['sorted']) . "'.split('|')";
 318
 319                 $encode = ($this->_encoding > 62) ? '_encode95' : $this->_getEncoder($ascii);
 320                 $encode = $this->_getJSFunction($encode);
 321                 $encode = preg_replace('/_encoding/','$ascii', $encode);
 322                 $encode = preg_replace('/arguments\\.callee/','$encode', $encode);
 323                 $inline = '\\$count' . ($ascii > 10 ? '.toString(\\$ascii)' : '');
 324
 325                 // $decode: code snippet to speed up decoding
 326                 if ($this->_fastDecode) {
 327                         // create the decoder
 328                         $decode = $this->_getJSFunction('_decodeBody');
 329                         if ($this->_encoding > 62)
 330                                 $decode = preg_replace('/\\\\w/', '[\\xa1-\\xff]', $decode);
 331                         // perform the encoding inline for lower ascii values
 332                         elseif ($ascii < 36)
 333                                 $decode = preg_replace($ENCODE, $inline, $decode);
 334                         // special case: when $count==0 there are no keywords. I want to keep
 335                         //  the basic shape of the unpacking funcion so i'll frig the code...
 336                         if ($count == 0)
 337                                 $decode = preg_replace($this->_safeRegExp('($count)\\s*=\\s*1'), '$1=0', $decode, 1);
 338                 }
 339
 340                 // boot function
 341                 $unpack = $this->_getJSFunction('_unpack');
 342                 if ($this->_fastDecode) {
 343                         // insert the decoder
 344                         $this->buffer = $decode;
 345                         $unpack = preg_replace_callback('/\\{/', array(&$this, '_insertFastDecode'), $unpack, 1);
 346                 }
 347                 $unpack = preg_replace('/"/', "'", $unpack);
 348                 if ($this->_encoding > 62) { // high-ascii
 349                         // get rid of the word-boundaries for regexp matches
 350                         $unpack = preg_replace('/\'\\\\\\\\b\'\s*\\+|\\+\s*\'\\\\\\\\b\'/', '', $unpack);
 351                 }
 352                 if ($ascii > 36 || $this->_encoding > 62 || $this->_fastDecode) {
 353                         // insert the encode function
 354                         $this->buffer = $encode;
 355                         $unpack = preg_replace_callback('/\\{/', array(&$this, '_insertFastEncode'), $unpack, 1);
 356                 } else {
 357                         // perform the encoding inline
 358                         $unpack = preg_replace($ENCODE, $inline, $unpack);
 359                 }
 360                 // pack the boot function too
 361                 $unpackPacker = new JavaScriptPacker($unpack, 0, false, true);
 362                 $unpack = $unpackPacker->pack();
 363
 364                 // arguments
 365                 $params = array($packed, $ascii, $count, $keywords);
 366                 if ($this->_fastDecode) {
 367                         $params[] = 0;
 368                         $params[] = '{}';
 369                 }
 370                 $params = implode(',', $params);
 371
 372                 // the whole thing
 373                 return 'eval(' . $unpack . '(' . $params . "))\n";
 374         }
 375
 376         var $buffer;
 377 // http://doc.spip.org/@_insertFastDecode
 378         function _insertFastDecode($match) {
 379                 return '{' . $this->buffer . ';';
 380         }
 381 // http://doc.spip.org/@_insertFastEncode
 382         function _insertFastEncode($match) {
 383                 return '{$encode=' . $this->buffer . ';';
 384         }
 385
 386         // mmm.. ..which one do i need ??
 387 // http://doc.spip.org/@_getEncoder
 388         function _getEncoder($ascii) {
 389                 return $ascii > 10 ? $ascii > 36 ? $ascii > 62 ?
 390                        '_encode95' : '_encode62' : '_encode36' : '_encode10';
 391         }
 392
 393         // zero encoding
 394         // characters: 0123456789
 395 // http://doc.spip.org/@_encode10
 396         function _encode10($charCode) {
 397                 return $charCode;
 398         }
 399
 400         // inherent base36 support
 401         // characters: 0123456789abcdefghijklmnopqrstuvwxyz
 402 // http://doc.spip.org/@_encode36
 403         function _encode36($charCode) {
 404                 return base_convert($charCode, 10, 36);
 405         }
 406
 407         // hitch a ride on base36 and add the upper case alpha characters
 408         // characters: 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ
 409 // http://doc.spip.org/@_encode62
 410         function _encode62($charCode) {
 411                 $res = '';
 412                 if ($charCode >= $this->_encoding) {
 413                         $res = $this->_encode62((int)($charCode / $this->_encoding));
 414                 }
 415                 $charCode = $charCode % $this->_encoding;
 416
 417                 if ($charCode > 35)
 418                         return $res . chr($charCode + 29);
 419                 else
 420                         return $res . base_convert($charCode, 10, 36);
 421         }
 422
 423         // use high-ascii values
 424         // characters: ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþ
 425 // http://doc.spip.org/@_encode95
 426         function _encode95($charCode) {
 427                 $res = '';
 428                 if ($charCode >= $this->_encoding)
 429                         $res = $this->_encode95($charCode / $this->_encoding);
 430
 431                 return $res . chr(($charCode % $this->_encoding) + 161);
 432         }
 433
 434 // http://doc.spip.org/@_safeRegExp
 435         function _safeRegExp($string) {
 436                 return '/'.preg_replace('/\$/', '\\\$', $string).'/';
 437         }
 438
 439 // http://doc.spip.org/@_encodePrivate
 440         function _encodePrivate($charCode) {
 441                 return "_" . $charCode;
 442         }
 443
 444         // protect characters used by the parser
 445 // http://doc.spip.org/@_escape
 446         function _escape($script) {
 447                 return preg_replace('/([\\\\\'])/', '\\\$1', $script);
 448         }
 449
 450         // protect high-ascii characters already in the script
 451 // http://doc.spip.org/@_escape95
 452         function _escape95($script) {
 453                 return preg_replace_callback(
 454                         '/[\\xa1-\\xff]/',
 455                         array(&$this, '_escape95Bis'),
 456                         $script
 457                 );
 458         }
 459 // http://doc.spip.org/@_escape95Bis
 460         function _escape95Bis($match) {
 461                 return '\x'.((string)dechex(ord($match)));
 462         }
 463
 464
 465 // http://doc.spip.org/@_getJSFunction
 466         function _getJSFunction($aName) {
 467                 $func = 'JSFUNCTION'.$aName;
 468                 if (isset($this->$func)){
 469                         return $this->$func;
 470                 }
 471                 else
 472                         return '';
 473         }
 474
 475         // JavaScript Functions used.
 476         // Note : In Dean's version, these functions are converted
 477         // with 'String(aFunctionName);'.
 478         // This internal conversion complete the original code, ex :
 479         // 'while (aBool) anAction();' is converted to
 480         // 'while (aBool) { anAction(); }'.
 481         // The JavaScript functions below are corrected.
 482
 483         // unpacking function - this is the boot strap function
 484         //  data extracted from this packing routine is passed to
 485         //  this function when decoded in the target
 486         // NOTE ! : without the ';' final.
 487         var $JSFUNCTION_unpack = 'function($packed, $ascii, $count, $keywords, $encode, $decode) {
 488     while ($count--) {
 489         if ($keywords[$count]) {
 490             $packed = $packed.replace(new RegExp(\'\\\\b\' + $encode($count) + \'\\\\b\', \'g\'), $keywords[$count]);
 491         }
 492     }
 493     return $packed;
 494 }';
 495 /*
 496 'function($packed, $ascii, $count, $keywords, $encode, $decode) {
 497     while ($count--)
 498         if ($keywords[$count])
 499             $packed = $packed.replace(new RegExp(\'\\\\b\' + $encode($count) + \'\\\\b\', \'g\'), $keywords[$count]);
 500     return $packed;
 501 }';
 502 */
 503
 504         // code-snippet inserted into the unpacker to speed up decoding
 505         var $JSFUNCTION_decodeBody = '    if (!\'\'.replace(/^/, String)) {
 506         // decode all the values we need
 507         while ($count--) {
 508             $decode[$encode($count)] = $keywords[$count] || $encode($count);
 509         }
 510         // global replacement function
 511         $keywords = [function ($encoded) {return $decode[$encoded]}];
 512         // generic match
 513         $encode = function () {return \'\\\\w+\'};
 514         // reset the loop counter -  we are now doing a global replace
 515         $count = 1;
 516     }
 517 ';
 518 //};
 519 /*
 520 '       if (!\'\'.replace(/^/, String)) {
 521         // decode all the values we need
 522         while ($count--) $decode[$encode($count)] = $keywords[$count] || $encode($count);
 523         // global replacement function
 524         $keywords = [function ($encoded) {return $decode[$encoded]}];
 525         // generic match
 526         $encode = function () {return\'\\\\w+\'};
 527         // reset the loop counter -  we are now doing a global replace
 528         $count = 1;
 529     }';
 530 */
 531
 532          // zero encoding
 533          // characters: 0123456789
 534          var $JSFUNCTION_encode10 = 'function($charCode) {
 535     return $charCode;
 536 }';//;';
 537
 538          // inherent base36 support
 539          // characters: 0123456789abcdefghijklmnopqrstuvwxyz
 540          var $JSFUNCTION_encode36 = 'function($charCode) {
 541     return $charCode.toString(36);
 542 }';//;';
 543
 544         // hitch a ride on base36 and add the upper case alpha characters
 545         // characters: 0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ
 546         var $JSFUNCTION_encode62 = 'function($charCode) {
 547     return ($charCode < _encoding ? \'\' : arguments.callee(parseInt($charCode / _encoding))) +
 548     (($charCode = $charCode % _encoding) > 35 ? String.fromCharCode($charCode + 29) : $charCode.toString(36));
 549 }';
 550
 551         // use high-ascii values
 552         // characters: ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿À�?ÂÃÄÅÆÇÈÉÊËÌ�?Î�?�?ÑÒÓÔÕÖ×ØÙÚÛÜ�?Þßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþ
 553         var $JSFUNCTION_encode95 = 'function($charCode) {
 554     return ($charCode < _encoding ? \'\' : arguments.callee($charCode / _encoding)) +
 555         String.fromCharCode($charCode % _encoding + 161);
 556 }';
 557
 558 }
 559
 560
 561 class ParseMaster {
 562         var $ignoreCase = false;
 563         var $escapeChar = '';
 564
 565         // constants
 566         var $EXPRESSION = 0;
 567         var $REPLACEMENT = 1;
 568         var $LENGTH = 2;
 569
 570         // used to determine nesting levels
 571         var $GROUPS = '/\\(/';//g
 572         var $SUB_REPLACE = '/\\$\\d/';
 573         var $INDEXED = '/^\\$\\d+$/';
 574         var $TRIM = '/([\'"])\\1\\.(.*)\\.\\1\\1$/';
 575         var $ESCAPE = '/\\\./';//g
 576         var $QUOTE = '/\'/';
 577         var $DELETED = '/\\x01[^\\x01]*\\x01/';//g
 578
 579 // http://doc.spip.org/@add
 580         function add($expression, $replacement = '') {
 581                 // count the number of sub-expressions
 582                 //  - add one because each pattern is itself a sub-expression
 583                 $length = 1 + preg_match_all($this->GROUPS, $this->_internalEscape((string)$expression), $out);
 584
 585                 // treat only strings $replacement
 586                 if (is_string($replacement)) {
 587                         // does the pattern deal with sub-expressions?
 588                         if (preg_match($this->SUB_REPLACE, $replacement)) {
 589                                 // a simple lookup? (e.g. "$2")
 590                                 if (preg_match($this->INDEXED, $replacement)) {
 591                                         // store the index (used for fast retrieval of matched strings)
 592                                         $replacement = (int)(substr($replacement, 1)) - 1;
 593                                 } else { // a complicated lookup (e.g. "Hello $2 $1")
 594                                         // build a function to do the lookup
 595                                         $quote = preg_match($this->QUOTE, $this->_internalEscape($replacement))
 596                                                  ? '"' : "'";
 597                                         $replacement = array(
 598                                                 'fn' => '_backReferences',
 599                                                 'data' => array(
 600                                                         'replacement' => $replacement,
 601                                                         'length' => $length,
 602                                                         'quote' => $quote
 603                                                 )
 604                                         );
 605                                 }
 606                         }
 607                 }
 608                 // pass the modified arguments
 609                 if (!empty($expression)) $this->_add($expression, $replacement, $length);
 610                 else $this->_add('/^$/', $replacement, $length);
 611         }
 612
 613 // http://doc.spip.org/@exec
 614         function exec($string) {
 615                 // execute the global replacement
 616                 $this->_escaped = array();
 617
 618                 // simulate the _patterns.toSTring of Dean
 619                 $regexp = '/';
 620                 foreach ($this->_patterns as $reg) {
 621                         $regexp .= '(' . substr($reg[$this->EXPRESSION], 1, -1) . ')|';
 622                 }
 623                 $regexp = substr($regexp, 0, -1) . '/';
 624                 $regexp .= ($this->ignoreCase) ? 'i' : '';
 625
 626                 $string = $this->_escape($string, $this->escapeChar);
 627                 $string = preg_replace_callback(
 628                         $regexp,
 629                         array(
 630                                 &$this,
 631                                 '_replacement'
 632                         ),
 633                         $string
 634                 );
 635                 $string = $this->_unescape($string, $this->escapeChar);
 636
 637                 return preg_replace($this->DELETED, '', $string);
 638         }
 639
 640 // http://doc.spip.org/@reset
 641         function reset() {
 642                 // clear the patterns collection so that this object may be re-used
 643                 $this->_patterns = array();
 644         }
 645
 646         // private
 647         var $_escaped = array();  // escaped characters
 648         var $_patterns = array(); // patterns stored by index
 649
 650         // create and add a new pattern to the patterns collection
 651 // http://doc.spip.org/@_add
 652         function _add() {
 653                 $arguments = func_get_args();
 654                 $this->_patterns[] = $arguments;
 655         }
 656
 657         // this is the global replace function (it's quite complicated)
 658 // http://doc.spip.org/@_replacement
 659         function _replacement($arguments) {
 660                 if (empty($arguments)) return '';
 661
 662                 $i = 1; $j = 0;
 663                 // loop through the patterns
 664                 while (isset($this->_patterns[$j])) {
 665                         $pattern = $this->_patterns[$j++];
 666                         // do we have a result?
 667                         if (isset($arguments[$i]) && ($arguments[$i] != '')) {
 668                                 $replacement = $pattern[$this->REPLACEMENT];
 669
 670                                 if (is_array($replacement) && isset($replacement['fn'])) {
 671
 672                                         if (isset($replacement['data'])) $this->buffer = $replacement['data'];
 673                                         return call_user_func(array(&$this, $replacement['fn']), $arguments, $i);
 674
 675                                 } elseif (is_int($replacement)) {
 676                                         return $arguments[$replacement + $i];
 677
 678                                 }
 679                                 $delete = ($this->escapeChar == '' ||
 680                                            strpos($arguments[$i], $this->escapeChar) === false)
 681                                         ? '' : "\x01" . $arguments[$i] . "\x01";
 682                                 return $delete . $replacement;
 683
 684                         // skip over references to sub-expressions
 685                         } else {
 686                                 $i += $pattern[$this->LENGTH];
 687                         }
 688                 }
 689         }
 690
 691 // http://doc.spip.org/@_backReferences
 692         function _backReferences($match, $offset) {
 693                 $replacement = $this->buffer['replacement'];
 694                 $quote = $this->buffer['quote'];
 695                 $i = $this->buffer['length'];
 696                 while ($i) {
 697                         $replacement = str_replace('$'.$i--, $match[$offset + $i], $replacement);
 698                 }
 699                 return $replacement;
 700         }
 701
 702 // http://doc.spip.org/@_replace_name
 703         function _replace_name($match, $offset){
 704                 $length = strlen($match[$offset + 2]);
 705                 $start = $length - max($length - strlen($match[$offset + 3]), 0);
 706                 return substr($match[$offset + 1], $start, $length) . $match[$offset + 4];
 707         }
 708
 709 // http://doc.spip.org/@_replace_encoded
 710         function _replace_encoded($match, $offset) {
 711                 return $this->buffer[$match[$offset]];
 712         }
 713
 714
 715         // php : we cannot pass additional data to preg_replace_callback,
 716         // and we cannot use &$this in create_function, so let's go to lower level
 717         var $buffer;
 718
 719         // encode escaped characters
 720 // http://doc.spip.org/@_escape
 721         function _escape($string, $escapeChar) {
 722                 if ($escapeChar) {
 723                         $this->buffer = $escapeChar;
 724                         return preg_replace_callback(
 725                                 '/\\' . $escapeChar . '(.)' .'/',
 726                                 array(&$this, '_escapeBis'),
 727                                 $string
 728                         );
 729
 730                 } else {
 731                         return $string;
 732                 }
 733         }
 734 // http://doc.spip.org/@_escapeBis
 735         function _escapeBis($match) {
 736                 $this->_escaped[] = $match[1];
 737                 return $this->buffer;
 738         }
 739
 740         // decode escaped characters
 741 // http://doc.spip.org/@_unescape
 742         function _unescape($string, $escapeChar) {
 743                 if ($escapeChar) {
 744                         $regexp = '/'.'\\'.$escapeChar.'/';
 745                         $this->buffer = array('escapeChar'=> $escapeChar, 'i' => 0);
 746                         return preg_replace_callback
 747                         (
 748                                 $regexp,
 749                                 array(&$this, '_unescapeBis'),
 750                                 $string
 751                         );
 752
 753                 } else {
 754                         return $string;
 755                 }
 756         }
 757 // http://doc.spip.org/@_unescapeBis
 758         function _unescapeBis() {
 759                 if (!empty($this->_escaped[$this->buffer['i']])) {
 760                          $temp = $this->_escaped[$this->buffer['i']];
 761                 } else {
 762                         $temp = '';
 763                 }
 764                 $this->buffer['i']++;
 765                 return $this->buffer['escapeChar'] . $temp;
 766         }
 767
 768 // http://doc.spip.org/@_internalEscape
 769         function _internalEscape($string) {
 770                 return preg_replace($this->ESCAPE, '', $string);
 771         }
 772 }
 773 ?>